Prognostic and Biologic Significance of Transfer RNA-Derived Small RNAs (tsRNAs) Expression in Younger Adult Patients (Pts) with Cytogenetically Normal Acute Myeloid Leukemia (CN-AML)

Introduction: tsRNAs constitute a novel class of small non-coding RNAs (~18-40 nucleotides), which are generated during stress-induced cleavage or the maturation processes of transfer RNAs (tRNAs). This class of RNA has been recently shown (Balatti et al. PNAS 2017;114:8071; Kim et al. Nature 2017;552:57) to be dysregulated in some forms of cancer (i.e., chronic lymphocytic leukemia, lung, colon, breast and ovarian cancer). However, to our knowledge, the prognostic value and biologic implications of tsRNA expression in AML pts have not been previously studied.

Aims: The aims of our study were to determine whether tsRNA expression associates with pretreatment characteristics and clinical outcome of younger [aged <60 years (y)] adults with de novo CN-AML, and to gain biological insights into the functional role of tsRNA expression in AML.

Methods: We conducted small RNAseq in a training set (n=208; median age, 48 y; range, 18-59 y) and a validation set (n=90; median age, 45 y; range 17-59 y) of younger adults with de novo CN-AML. None of the clinical or molecular parameters analyzed in this study were significantly different between the two sets. All pts were treated on frontline Cancer and Leukemia Group B/Alliance protocols. To obtain the expression of tsRNAs, we applied a bioinformatics workflow to small RNAseq data of a cohort of 20 CN-AML pts, in association with data, which were collected via custom array assay (Pekarsky et al. PNAS 2016;113:5071).

Results: We obtained the expression data of 136 tsRNAs in both sets. Next, we derived a signature (3-tsRNA) composed of 3 tsRNAs (tsRNA20, tsRNA64, and tsRNA66), which were associated with event-free survival at the significance level of P<.001. We used the 3rd quartile as a cutoff in each set to discriminate between pts with a 3-tsRNAlow (1st to 3rd quartile) and those with 3-tsRNAhigh (4th quartile) score. With respect to clinical outcome, pts with 3-tsRNAhigh score had a worse disease-free survival (DFS; 5-y rates, 23% v 43%; P=.01) and overall survival (OS; 5-y rates, 25% v 45%; P=.003) in the training set (Figure 1). These results were confirmed in the validation set (5-y DFS rates, 12% v 50%; P=.003; 5-y OS rates, 18% v 58%; P=.002). In multivariable analyses, 3-tsRNAhigh score independently associated with shorter DFS (P=.02; HR: 2.34) and OS (P=.03; HR: 1.98) after adjusting for other co-variates [i.e., internal tandem duplication of the FLT3 gene (FLT3-ITD) and MN1 expression].

Regarding pretreatment clinical and molecular features in the training set, pts with a 3-tsRNAhigh score had a higher percent of bone marrow blasts (P=.03) and tended to have a higher percent of blood blasts (P=.06). 3-tsRNAhigh scorers had a higher frequency of FLT3-ITD (P=.01) and DNMT3A non-R882 mutations (P=.03), and were marginally more likely to have an ASXL1 (P=.06) or RUNX1 mutation (P=.06). Pts with 3-tsRNAhigh score were less likely to harbor NPM1 mutations (P=.03).

To gain initial insights into the biological significance of tsRNA expression in AML, we performed loss of function experiments and knocked-down (KD) all 3 tsRNAs using customized single strand RNA-inhibitors. Of the tsRNAs tested, only KD of tsRNA20 and tsRNA66 decreased the proliferative capacity of THP-1 and OCI-AML3 cells, as measured by MTS reagent degradation (tsRNA20: P=.002 and P=.04, respectively; tsRNA66: P=.009 and P=.07, respectively). Additionally, these cell lines showed higher frequency of apoptotic cells 48 hours after KD (tsRNA20: P=.007 and P=.008, respectively; tsRNA66: P<.001 and P=.009, respectively).

Conclusion: We conclude that tsRNAs represent a novel prognostic and biologically important class of non-coding RNAs in CN-AML.

View largeDownload slide

View largeDownload slide

Close modal